---

Phi-4-reasoning-plus-GGUF

Original Model

microsoft/Phi-4-reasoning-plus

Run with LlamaEdge

• LlamaEdge version: coming soon

• Prompt template

  • Prompt type: phi-4-chat

  • Prompt string

    <|im_start|>system<|im_sep|>
    {system_message}<|im_end|>
    <|im_start|>user<|im_sep|>
    {user_message}<|im_end|>
    <|im_start|>assistant<|im_sep|>
    

• Context size: 32000

• Run as LlamaEdge service

  wasmedge --dir .:. --nn-preload default:GGML:AUTO:Phi-4-reasoning-plus-Q5_K_M.gguf \
    llama-api-server.wasm \
    --prompt-template phi-4-chat \
    --ctx-size 32000 \
    --model-name Phi-4-reasoning-plus
  

• Run as LlamaEdge command app

  wasmedge --dir .:. --nn-preload default:GGML:AUTO:Phi-4-reasoning-plus-Q5_K_M.gguf \
    llama-chat.wasm \
    --prompt-template phi-4-chat \
    --ctx-size 32000
  

Quantized GGUF Models

Name	Quant method	Bits	Size	Use case
Phi-4-reasoning-plus-Q2_K.gguf	Q2_K	2	5.55 GB	smallest, significant quality loss - not recommended for most purposes
Phi-4-reasoning-plus-Q3_K_L.gguf	Q3_K_L	3	7.93 GB	small, substantial quality loss
Phi-4-reasoning-plus-Q3_K_M.gguf	Q3_K_M	3	7.36 GB	very small, high quality loss
Phi-4-reasoning-plus-Q3_K_S.gguf	Q3_K_S	3	6.50 GB	very small, high quality loss
Phi-4-reasoning-plus-Q4_0.gguf	Q4_0	4	8.38 GB	legacy; small, very high quality loss - prefer using Q3_K_M
Phi-4-reasoning-plus-Q4_K_M.gguf	Q4_K_M	4	9.05 GB	medium, balanced quality - recommended
Phi-4-reasoning-plus-Q4_K_S.gguf	Q4_K_S	4	8.44 GB	small, greater quality loss
Phi-4-reasoning-plus-Q5_0.gguf	Q5_0	5	10.2 GB	legacy; medium, balanced quality - prefer using Q4_K_M
Phi-4-reasoning-plus-Q5_K_M.gguf	Q5_K_M	5	10.6 GB	large, very low quality loss - recommended
Phi-4-reasoning-plus-Q5_K_S.gguf	Q5_K_S	5	10.2 GB	large, low quality loss - recommended
Phi-4-reasoning-plus-Q6_K.gguf	Q6_K	6	12.0 GB	very large, extremely low quality loss
Phi-4-reasoning-plus-Q8_0.gguf	Q8_0	8	15.6 GB	very large, extremely low quality loss - not recommended
Phi-4-reasoning-plus-f16.gguf	f16	16	29.3 GB

Quantized with llama.cpp b5201.